WO2020238954A1

WO2020238954A1 - Apnea monitoring method and device

Info

Publication number: WO2020238954A1
Application number: PCT/CN2020/092597
Authority: WO
Inventors: 黄晓萍
Original assignee: 华为技术有限公司
Priority date: 2019-05-31
Filing date: 2020-05-27
Publication date: 2020-12-03
Also published as: EP3954278A4; US20220225930A1; CN110301890A; CN110301890B; EP3954278A1

Abstract

An apnea monitoring method and device, relating to the field of medical monitoring. The apnea monitoring method comprises: collecting first sound of at least one user by a first terminal, and using directional beamforming of a multi-microphone array configured by the first terminal to determine the first sound corresponding to the users in different beam coverage areas, or distinguishing the first sound of different users by means of a voiceprint recognition algorithm; obtaining a power spectral density corresponding to the first sound according to the first sound; and if the power spectral density is lower than a first threshold, determining that the user has apnea. By means of the apnea monitoring method, apnea monitoring on one or more users may be achieved without wearing a monitoring instrument. The operation is simple and flexible, and the monitoring experience of the user is improved.

Description

Method and device for apnea monitoring

This application claims the priority of the Chinese patent application filed with the State Intellectual Property Office of China with the application number 201910471804.2, and the priority of the Chinese patent application with the title of "Method and Apparatus for Apnea Monitoring" on May 31, 2019. The entire content is incorporated into this application by reference.

Technical field

This application relates to the field of medical monitoring, and more specifically to a method and device for apnea monitoring.

Background technique

With the current increasing social pressure, overweight and irregular life and other factors, more than 20% of people have symptoms of sleep apnea or hypopnea. In daily life, it is difficult for users with sleep-disordered breathing to perceive the time and severity of apnea on their own initiative, which may result in the failure of timely treatment and improvement of the symptoms, thereby posing a great threat to the user's health.

At present, a variety of sleep breathing monitoring instruments can be used to monitor the sleep process, so as to help the user know whether there is apnea disorder during sleep. Commonly used sleep apnea monitoring methods include the use of polysomnography (PSG) to diagnose obstructive sleep apnea, use a home sleep apnea screener to monitor the user’s obstructive sleep apnea symptoms or use an oximeter to monitor the user Changes in blood oxygen during sleep to assist in monitoring the user's apnea status. However, the currently used instruments all need to be worn by the user, which has poor comfort, or needs to be monitored in a hospital, and it takes a long time to obtain reports.

Summary of the invention

The embodiments of the present application provide a method and device for apnea monitoring to solve the problem of poor comfort caused by the need to wear an instrument in the process of sleep respiration monitoring, and the inability to achieve simultaneous sleep respiration monitoring by multiple people.

In a first aspect, a method for apnea monitoring is provided, including: collecting the first sound of at least one user through a multi-microphone array of a first terminal, wherein each user is located in a different beam coverage of the multi-microphone array Range; obtain the power spectral density corresponding to the first sound according to the first sound; when the power spectral density is lower than a first threshold, determine that the user has apnea.

It should be understood that the multi-microphone array can have directional beam forming by setting. When the user is located within the coverage of the directional beam of the multi-microphone array, the microphone can collect the good audio signal of the user’s first voice, so that The audio signal is subsequently used to analyze the user's sleep apnea status.

With reference to the first aspect, in some implementations of the first aspect, the collecting the first voice of at least one user through the multi-microphone array of the first terminal includes: according to the location of the first terminal and the user, The sound pickup parameters of the multi-microphone array are set so that different beams of the multi-microphone array cover different users; the multi-microphone array collects first sounds of the users in different beam coverage areas.

According to the method for apnea monitoring provided by the embodiment of the present application, the first sound collection is performed on the users located in the directional beam coverage of different multi-microphones through the multi-microphone array of the first terminal. The beam corresponding to a sound source does distinguish the first sounds of different users, and then the apnea status of the user is determined according to the first sounds of different users.

With reference to the first aspect, in some implementations of the first aspect, the obtaining the power spectral density corresponding to the first sound according to the first sound includes: obtaining the first sound according to the first sound The time-domain signal; the time-domain signal is Fourier transformed to obtain the power spectral density of the first sound.

It should be understood that the microphone of the first terminal can obtain the time domain signal of the first sound according to the collected first sound. In order to facilitate intuitive analysis of the audio signal of the first sound, the time domain signal can be converted into Power spectral density.

With reference to the first aspect, in some implementations of the first aspect, the method further includes: pre-collecting environmental noise of the environment where the user is located, and determining the frequency bandwidth of the environmental noise; Frequency bandwidth, filtering the environmental noise.

According to the method for apnea monitoring provided by the embodiments of the present application, by filtering the environmental noise of the environment where the monitored user is located, the user’s sleep sound can be better obtained, where the sleep sound can be breathing sound or snoring sound to reduce environmental noise Interference to the user's sleep sound analysis process.

With reference to the first aspect, in some implementations of the first aspect, the method further includes: using a voiceprint recognition algorithm to determine first voices corresponding to different users.

It should be understood that when multiple users are simultaneously monitored for sleep apnea, the voiceprint recognition algorithm can be used to more accurately distinguish the first voices of different users, so as to facilitate subsequent analysis of the sleep apnea conditions of different users.

With reference to the first aspect, in some implementations of the first aspect, the method further includes: determining the sleep duration of the user; and determining according to the number of times the apnea occurs during the sleep of the user and the sleep duration The user's breathing disorder AHI index.

According to the method for apnea monitoring provided by the embodiments of the present application, the user’s AHI index is determined according to the number of apneas and sleep duration during a complete sleep of the user, so as to more intuitively and accurately determine the severity of the user’s apnea degree.

In a second aspect, a method for monitoring apnea is provided, including: collecting first sounds of multiple users through a first terminal; determining first sounds corresponding to different users according to a voiceprint recognition algorithm; A sound acquires the power spectral density corresponding to the first sound; when the power spectral density is lower than a first threshold, it is determined that the user has apnea.

According to the method for apnea monitoring provided by the embodiments of the present application, when multiple users are simultaneously monitored for sleep apnea, the voiceprint recognition algorithm can be used to more accurately distinguish the first voices of different users, so as to more accurately target different users’ voices. Analysis of sleep apnea status.

With reference to the second aspect, in some implementations of the second aspect, the first terminal has a multi-microphone array; the collecting the first sounds of multiple users through the first terminal includes: using the multi-microphone of the first terminal The array collects the first sounds of multiple users, where each user is located in a different beam coverage area of the multi-microphone array.

With reference to the second aspect, in some implementations of the second aspect, the collecting the first sounds of multiple users through the multi-microphone array of the first terminal includes: according to the positions of the first terminal and the user, The sound pickup parameters of the multi-microphone array are set so that different beams of the multi-microphone array cover different users; the multi-microphone array collects the first sounds of the users in different beam coverage areas.

According to the method for apnea monitoring provided by the embodiments of the present application, the first sound collection is performed on users located within the coverage of the directional beams of different multi-microphones through the multi-microphone array of the first terminal. The beam corresponding to the source does distinguish the first voices of different users, and then the apnea status of the user is determined according to the voices of different users. Through the method of this embodiment, the sleep apnea monitoring of one user or multiple users can be realized without the user wearing a monitoring instrument.

With reference to the second aspect, in some implementations of the second aspect, the obtaining the power spectral density corresponding to the first sound according to the first sound includes: obtaining the first sound according to the first sound The time-domain signal; the time-domain signal is Fourier transformed to obtain the power spectral density of the first sound.

It should be understood that the microphone can obtain the time domain signal of the first sound according to the collected first sound. In order to facilitate intuitive analysis of the audio signal of the first sound, the time domain signal can be converted into power spectral density by Fourier transform.

With reference to the second aspect, in some implementations of the second aspect, the method further includes: determining the sleep duration of the user; determining according to the number of times the apnea occurs during the sleep of the user and the sleep duration The AHI index of the user.

According to the apnea monitoring method provided by the embodiments of the present application, by filtering the environmental noise of the environment where the monitored user is located, the user's sleep sound can be better obtained, where the sleep sound can be breathing sound or snoring sound to reduce environmental noise Interference to the user's sleep sound analysis process.

In a third aspect, an apparatus for detecting apnea is provided, including: a sound collecting unit for collecting a first sound of at least one user; and a data processing unit for acquiring information of the first sound according to the first sound Power spectral density; the data processing unit is further configured to determine that the user has apnea when the power spectral density is lower than a first threshold.

With reference to the third aspect, in some implementations of the third aspect, the obtaining the power spectral density of the first sound according to the first sound includes: obtaining the power spectral density of the first sound according to the first sound Time domain signal; the time domain signal is Fourier transformed to obtain the power spectral density of the first sound.

With reference to the third aspect, in some implementations of the third aspect, the sound collection unit is further configured to collect in advance the environmental noise of the environment where the user is located; the data processing unit is configured to determine the environmental noise According to the frequency bandwidth of the environmental noise, the environmental noise is filtered.

With reference to the third aspect, in some implementation manners of the third aspect, the data processing unit is further configured to use a voiceprint recognition algorithm to determine first voices corresponding to different users.

With reference to the third aspect, in some implementations of the third aspect, the data processing unit is further configured to determine the sleep duration of the user, and according to the number of apneas and the total number of apneas that occur during the sleep of the user. The sleep duration determines the AHI index of the user.

In a fourth aspect, an apparatus for apnea monitoring is provided, including at least one microphone and a processor, wherein the apparatus is used to execute the method according to any one of the implementation manners of the first aspect and the second aspect.

According to the method and device for apnea monitoring provided by the embodiments of the present application, by using the directional multi-beam forming of a multi-microphone array or voiceprint recognition to distinguish the voice information of different users, the apnea conditions of multiple users can be monitored simultaneously, and the users There is no need to wear equipment and will not affect sleep.

Description of the drawings

Figure 1 is a possible application scenario of the apnea monitoring method provided by this application.

Fig. 2 is a schematic flowchart of a method for apnea monitoring provided by an embodiment of the application.

FIG. 3 is a schematic flowchart of another method for monitoring apnea provided by an embodiment of the application.

FIG. 4 is a schematic flowchart of another method for monitoring apnea according to an embodiment of the application.

FIG. 5 is a schematic flowchart of another method for monitoring apnea according to an embodiment of the application.

Fig. 6 is a schematic structural diagram of an apnea monitoring device provided by an embodiment of the application.

Fig. 7 is a schematic structural diagram of another apnea monitoring device provided by an embodiment of the application.

Detailed ways

The technical solution in this application will be described below in conjunction with the drawings.

Obstructive sleep apnea hypopnea syndrome (OSAHS), generally refers to an adult’s normal sleep time (e.g. 7 hours) every night, the mouth and nose have stopped airflow for 10 seconds or more for more than 30 times. Times or more. At present, the traditional screening and diagnosis methods for obstructive sleep apnea include polysomnography (PSG) or the use of household sleep apnea screeners, blood oximeters and other equipment for monitoring. The following is a brief introduction to the currently adopted user sleep monitoring methods.

1. Use polysomnography to monitor PSG to diagnose obstructive sleep apnea.

PSG is the "gold standard" for diagnosing sleep apnea syndrome. While monitoring obstructive sleep apnea, it can also identify many sleep-related diseases. The monitoring process is mainly that during the examination, the doctor pastes some electrodes on the examinee and connects them to the instrument. By detecting the user's nose and mouth airflow, snoring, chest and abdomen movements, it is judged whether there is apnea or hypoventilation. It happened. In addition, PSG can further obtain the maximum and average duration of apneas to distinguish between central, obstructive, or mixed apneas. Because PSG needs to use a specific PSG instrument to monitor the obstructive sleep apnea of users, and the monitoring is cumbersome, the detection location environment is too restrictive, and it causes a poor sleep experience for the user, which is difficult to achieve in daily life. Convenient monitoring.

2. Use a home sleep apnea screener to monitor the user's sleep.

In view of the complexity of PSG and the inconvenience of use, there is still a portable sleep apnea screener that retains the core monitoring capabilities of PSG, but has a miniaturized structure. The principle of the home sleep apnea screener to monitor whether the user has apnea is similar to that of PSG. Since it also needs to be worn by the user, it will also cause a poor user experience. In addition, it may not achieve the effect of simultaneously monitoring multiple people with the same instrument.

3. The oximeter assists in screening sleep apnea.

Since the obstructive apnea of the user will cause the blood oxygen to drop, the blood oxygen of the user can be monitored during sleep to achieve the purpose of assisting in the screening of sleep apnea syndrome. However, in the actual application process, the drop in blood oxygen may be caused by a variety of reasons, such as heart failure. The method of assisted monitoring by this oximeter cannot accurately identify the cause of the drop in blood oxygen, which leads to large errors in the monitoring results. .

In view of the deficiencies in the current monitoring methods for obstructive sleep apnea syndrome of users, the embodiments of the present application provide a method that does not need to be worn by the user and can simultaneously monitor the sleep of multiple people through one device. The sleep monitoring method provided in the present application will be introduced below in conjunction with the drawings.

Figure 1 shows a possible application scenario of the apnea monitoring method provided by this application.

First, it should be understood that the microphone can have an omnidirectional pickup response, that is, the microphone can respond to sounds from all directions. A multi-microphone array composed of multiple microphones can form a directional response or beam field pattern. After design, the microphone array can be more sensitive to sounds from one or more specific directions, and can collect better sound signals in that specific direction.

The sleep monitoring method provided by the embodiments of the present application uses the beamforming technology of the multi-microphone array to collect the audio signals of the sleep sounds of users located in different positions, and the sleep process of multiple users can be performed simultaneously without the user wearing an instrument. Monitor, and then determine whether the multiple users have obstructive sleep pause or the severity of obstructive sleep pause. Among them, the sleep sound may be, for example, the breathing sound and snoring sound of the user during sleep. In addition, the multi-microphone array may be an existing multi-microphone array in the existing equipment, that is, an existing terminal device equipped with a multi-microphone array can be used to collect sleep sounds of multiple users, such as a mobile phone or a recording device. Pen etc.

Fig. 2 shows a schematic flowchart of a method for apnea monitoring provided by an embodiment of the present application. The method includes the following steps:

S210: Collect a first voice of at least one user.

Wherein, a first terminal configured with a microphone may be used to collect the first voice of the monitored user, where the first terminal may be, for example, a mobile phone or other recording device configured with a single microphone or multiple microphone arrays; The sound is the sleep sound of the user during sleep, for example, it may be the breathing sound and snoring sound of the user during sleep.

It should be understood that the method for monitoring apnea provided by the embodiments of the present application can be used to monitor the sleep and breathing status of multiple users at the same time. Therefore, when the sleep sounds of multiple users are collected, in order to realize the subsequent sleep of each user. To analyze the apnea status, it is necessary to identify the sleep sounds corresponding to different users. Among them, the embodiment of the present application mainly adopts two methods to distinguish the sleep sounds of different users: First, the first terminal equipped with a multi-microphone array is used to collect the first sounds of multiple users, and the beam is formed according to the directional beam of the multi-microphone array. Collect the first voices of users located in different beam ranges, and then determine the corresponding relationship between the user and the first voice according to the source of the first voice and the location of the user; Second, collect the first voices of multiple users through the first terminal Then, according to the voiceprint recognition algorithm, the first voice of different users is determined.

As an example, a first device configured with a multi-microphone array is used to record the first voice of at least one user. It should be understood that by setting the pickup parameters of the microphone array, such as pickup direction, pickup angle width, etc., combined with the position of the first device and the user, it can be used within the beam coverage of the multi-microphone array. In addition, by adjusting the phase of the signal of the microphone array unit, the phase-adjusted signals of each unit are superimposed to obtain the main lobe signal in the direction specified by the user, so that the first device can accurately collect the first voice of the user at different positions .

It should be understood that when monitoring the sleep breathing process of multiple users at the same time, since different users are located in the coverage areas of different beams of the multi-microphone array, the first sound can be determined according to the beam coverage corresponding to the source of the first sound. Correspondence between sound and user. That is, when multiple first sounds of multiple users are collected, the first voice corresponding to each user can be determined according to the beam coverage of the multiple users and the beam corresponding to the source of the multiple first sounds. One voice to facilitate subsequent analysis of the apnea conditions of different users’ first voices.

Optionally, the first device configured with a multi-microphone array may be a mobile phone, a voice recorder or other terminal devices with a multi-microphone array, which is not limited in this application.

As another example, after the first voice of the user is collected by the first device, the first voice of different users is determined by a voiceprint recognition algorithm. It should be understood that voiceprint is a sound wave spectrum that carries speech information displayed by an electroacoustic instrument. Because people’s vocal organs differ greatly in size and shape, the voiceprint atlas of any two people are not the same. Under normal circumstances, different people can be distinguished by voiceprint information or to determine whether the collected sound belongs to The voice of the same person. In this embodiment, the process of determining the first voice corresponding to the user by means of voiceprint recognition may be, for example, in the process of performing a formal sleep breathing monitoring of the monitored user, collecting the first voice of each user, and obtaining short voices of different users. Time speech spectrum; extract the voiceprint feature according to the short-term speech spectrum; determine the user corresponding to the first voice through the voiceprint recognition model and the user's voiceprint feature. Among them, the voiceprint recognition model can be an existing model, such as a Gaussian mixture model (GMM), a support vector machine (SVM) model, a channel model, an identity vector (i-vector) ) Model etc.

Among them, the GMM model is taken as an example to further introduce the voiceprint recognition process.

It should be understood that the difference in voices of different people is mainly reflected in the difference in the short-term speech spectrum, and the difference in the short-term speech spectrum can be measured by its probability density function. The GMM model can fit the probability density of the spatial distribution with the weighted sum of multiple Gaussian probability density functions. The fitted Gaussian probability density function can smoothly approximate the probability density function of any shape, and is an easy-to-handle parameter model . In terms of specific representation, the parameter model may be a model in which the mean vector of each Gaussian component of the GMM model is arranged together to form a super vector as a certain speaker.

Exemplarily, the voiceprint recognition model involved in this embodiment may be a GMM model based on Mel frequency cepstrum coefficient (MFCC). The process of voiceprint recognition through the GMM model can be: (1) GMM model training. In the GMM training process, multiple voice signals can be collected, and the feature parameters of the voice signal can be extracted after preprocessing the multiple voice signals. Among them, the voice feature parameter can be, for example, MFCC; use the feature parameter of the voice signal as a sample training GMM model. (2) Voiceprint recognition process: Obtain the characteristic parameters of the first sound according to the collected first sound, and compare the characteristic parameters of the first sound with the established GMM model, and determine the characteristic parameters of the first sound according to the recognition accuracy rate of the GMM The user corresponding to the first voice.

It should be understood that the voiceprint recognition process used in the embodiments of the present application can also refer to other existing processes, and is not limited to the voiceprint recognition method mentioned above.

It should be understood that in the actual process of recording the user's sleep sound, environmental noise is unavoidable. Therefore, in order to obtain an accurate and good sleep sound signal in the subsequent analysis process, the environmental noise needs to be eliminated.

Exemplarily, the process of eliminating or reducing environmental noise in the embodiment of the present application may be: before the user’s sleep sound signal is formally collected, the environmental noise of the environment where the user is located is first collected to obtain the sound wave corresponding to the environmental noise. In the follow-up process of sound recording for insomnia monitoring of users, the multi-microphone generates a sound wave with the opposite phase and the same frequency and amplitude as the pre-recorded environmental noise sound wave, so that it interferes with the environmental noise. Phase cancellation, thereby eliminating environmental noise; or, according to the frequency bandwidth of the pre-recorded environmental noise, set the filter frequency bandwidth of the filter so that the filter can filter the noise during the recording process. Among them, the method of filtering environmental noise can also be other existing methods, and the application is not limited to this.

It should be understood that the first sound of the user during sleep is recorded through the microphone, and then the user’s sleep apnea is analyzed, so that the user can monitor sleep apnea without wearing a monitoring instrument, which improves the user’s monitoring Comfort in the process.

Step S220: Acquire the power spectral density corresponding to the first sound according to the first sound.

Generally speaking, the sound signal obtained through the microphone is a time domain signal. In order to analyze the user's sound signal more intuitively, the time domain signal is converted into a frequency domain signal. Specifically, the obtained time-domain signals of sleep sounds of different users are obtained through Fourier transform to obtain their corresponding power spectra, that is, power spectral density. Among them, the formula used in the Fourier transform process is:

Among them, ω is frequency, t is time, e- ^iωt is a complex variable function, f(t) can be the time domain signal of the first sound, and F(ω) can be the frequency domain signal after Fourier transform.

S230: When the power spectral density is lower than the first threshold, it is determined that the user has an apnea.

Wherein, when the power spectral density of the sleep sound of the user is lower than the first threshold, it is determined that the user has apnea.

It should be understood that the first threshold may be a value set according to the user's normal breathing situation. Exemplarily, the process of obtaining the first threshold may be: performing time-frequency conversion according to the time-frequency signal of the sleep sound in the user's normal breathing state (for example, a breathing state in which no apnea has occurred just after falling asleep) to obtain the breathing sound in this state And take the average value of the respiratory spectrum peaks of multiple consecutive times (such as 5 to 10 times), and then take 30% to 50% of the average value as the first threshold to determine whether the user has apnea and breathing. The severity of the suspension. In addition, the first threshold can be optimized and adjusted based on clinical test data and with reference to the PSG standard accuracy to achieve the best sensitivity and specificity.

Optionally, when the power spectral density of the user's sleep sound is lower than the first threshold for a duration that reaches the second threshold, it is determined that the user has an apnea, where the second threshold may be, for example, 10s.

Optionally, the respiratory disorder (apnea-hypopnea index, AHI) index of the user is determined according to the number of times the power spectral density of the first sound of the different user is lower than the first threshold. Specifically, the AHI index of the user may be determined according to the number of times the power spectral density of the first sound is lower than the first threshold and the sleep duration of the user.

Optionally, when monitoring the user's sleep, the user's sleep time may be recorded. For example, the sleep time may be the sleep duration of the user throughout the night.

As an example, the AHI index of the user is determined according to the power spectral density of the user's breathing sound and/or the power spectral density of the snoring sound and the sleep time of the user. Exemplarily, when the power spectral density of the user’s breathing sound or the power spectral density of the snoring sound is lower than the first threshold, and the length of time lower than the first threshold reaches the second threshold, it is recorded as an apnea. During the user's entire sleep , Record the total number of apneas (denoted as k), and calculate the sleep-disordered breathing coefficient according to its entire sleep duration (denoted as t), where AHI=k/t; or record the user’s sleep sound per unit time The power spectral density of is lower than the first threshold, and the number of times that the length of time lower than the first threshold reaches the second threshold.

Among them, the severity of the apnea of the user is determined according to the user's AHI index. Exemplarily, when the AHI is in the range less than or equal to 5, it is determined that the user sleeps normally, that is, no apnea is considered; when the AHI value is in the range of 5 to 15, the user is considered to have mild Sleep apnea status; when the AHI value falls within the range of 15 to 30, the user is considered to have moderate sleep apnea status; when the AHI is within the range greater than or equal to 30, the user is considered to have severe sleep Apnea condition.

According to the method for apnea monitoring provided by the present application, by using the directional multi-beam forming of a multi-microphone array or voiceprint recognition to distinguish the voice information of different users, the apnea status of multiple users can be monitored simultaneously, and the user does not need to wear an instrument , Will not affect sleep.

The method for monitoring apnea of the present application will be specifically introduced below in conjunction with the accompanying drawings.

Fig. 3 shows a schematic flowchart of a method for apnea monitoring provided by an embodiment of the present application. It includes the following steps:

S310. Use the directional beam forming of the multi-microphone array to collect first voices of users in different positions.

Wherein, a first terminal configured with a microphone may be used to collect the first voice of the monitored user. The first terminal may be, for example, a mobile phone or other recording device configured with a single microphone or multiple microphone arrays; the first voice is The sleep sound of the user during sleep may be, for example, the breathing sound and snoring sound of the user during sleep.

Wherein, multiple users to be monitored may be respectively located within the coverage of the directional beam of the multi-microphone array.

Optionally, before the user's sleep breathing is monitored, the sleep apnea monitoring function of the first device is turned on manually or in other ways.

Optionally, after the first terminal collects the first voices of multiple users, it may also recognize the first voices of different users through a voiceprint recognition algorithm, so as to more accurately distinguish the first voices of different users.

As an example, the process of recording the first voice of at least one user with a first device configured with a multi-microphone array may include: setting the pickup parameters of the microphone array, such as pickup direction, pickup angle width, etc., combined The position of the first device and the user can be used within the beam coverage of the multi-microphone array; in addition, by adjusting the phase of the signal of the microphone array unit, the phase-adjusted signals of each unit are superimposed to obtain the user-specified direction The main lobe signal, so that the first device accurately collects the first sounds of users located in different positions.

It should be understood that in the process of actually recording the user's sleep sound, there will inevitably be environmental noise. Therefore, in order to obtain an accurate and good sleep sound signal in the subsequent analysis process, the environmental noise needs to be eliminated or reduced.

S320: Acquire a power spectral density of the first sound according to the first sound of the user.

Wherein, the multi-microphone array can obtain the time domain signal of the first sound according to the collected first sound.

Optionally, the Fourier transform can be used to convert the signal in the time domain into the power spectral density in the frequency domain. For the specific conversion process, refer to step S220, and to avoid repetition, it will not be repeated here.

S330: When the power spectral density is lower than the first threshold, it is determined that the user has an apnea.

Optionally, when the power spectral density of the user's sleep sound is lower than the first threshold and reaches the second threshold, it is determined that the user has an apnea, where the second threshold may be, for example, 10s or 15s.

Optionally, according to the number of times the power spectral density of the first voice of the different user is lower than the first threshold, the respiratory disorder AHI index of the user is determined.

Optionally, according to the power spectral density of the first voices of different users, the respiratory disorder AHI index of the user is determined. Specifically, the AHI index of the user may be determined according to the number of times the power spectral density of the first sound is lower than the first threshold and the sleep duration of the user.

As an example, when the user’s breath sound power spectral density or snoring sound power spectral density is lower than the first threshold, and the length of time lower than the first threshold reaches the second threshold, it is recorded as an apnea. , Record the total number of apneas (denoted as k), and calculate the sleep-disordered breathing coefficient according to the entire sleep duration (denoted as t), where AHI=k/t; or, record the user’s sleep per unit time The number of times that the power spectral density of the sound is lower than the first threshold, and the length of time lower than the first threshold reaches the second threshold.

It should be understood that after obtaining the user's AHI index, the user's apnea level can be judged according to the AHI index. For example, when the AHI is in the range less than or equal to 5, it is determined that the user sleeps normally, that is, no apnea is considered; when the AHI value is in the range of 5 to 15, the user is considered to have a light sleep Apnea condition; when the value of AHI is in the range of 15 to 30, the user is considered to have moderate sleep apnea; when the AHI is in the range greater than or equal to 30, the user is considered to have severe sleep apnea situation.

According to the method for apnea provided by the embodiment of the present application, the sleep sound of a user is collected through a multi-microphone array, which can realize sleep apnea monitoring and risk assessment for multiple users in a non-contact manner.

FIG. 4 shows a schematic flowchart of another method for monitoring apnea provided by an embodiment of the present application. It includes the following steps:

S410: Collect first sounds of multiple users through the first terminal.

Optionally, the first terminal may be a mobile phone, a voice recorder, etc., and may also be other terminal devices configured with a single or multiple microphones.

Optionally, the first sound of the user is recorded through a microphone in the first device, where the first sound is breathing sound, snoring sound, etc. of the user during sleep.

Optionally, the microphone in the first terminal may obtain the time domain signal of the first sound according to the collected first sound.

It should be understood that in the actual process of recording the user's sleep sound, environmental noise is unavoidable. Therefore, in order to obtain an accurate and good sleep sound signal in the subsequent analysis process, the environmental noise needs to be eliminated or reduced.

Exemplarily, the process of eliminating or reducing environmental noise in the embodiment of the present application may be: before the user’s sleep sound signal is formally collected, the environmental noise of the environment where the user is located is first collected to obtain the sound wave corresponding to the environmental noise. In the subsequent sound recording process of the insomnia monitoring of the user, the multi-microphone generates a sound wave with the opposite phase and the same frequency and amplitude as the pre-recorded environmental noise sound wave, so that it interferes with the environmental noise to achieve Phase cancellation to eliminate environmental noise; or, according to the frequency bandwidth of the pre-recorded environmental noise, set the filter frequency bandwidth of the filter so that the filter can filter the noise during the recording process. Among them, the method of filtering environmental noise can also be other existing methods, and the application is not limited to this.

S420: Determine the first voice corresponding to the user according to the voiceprint recognition algorithm.

Optionally, before collecting the first sound during the user’s sleep, that is, in the pre-processing stage, pre-collect the user’s breathing sound or snoring when sleep apnea does not occur. For example, the user’s muscles may not be in the pre-sleep phase. In a fully relaxed state and no sleep apnea, a segment of the user's breathing and/or snoring sounds are collected.

Optionally, the voice feature of the user's first voice is extracted, and voiceprint models of different users are established.

Optionally, the first voice of different users is determined according to the first voice obtained in the formal monitoring process and the voiceprint models of different users.

As an example, after the user's first voice is collected by the first device, the voiceprint recognition algorithm is used to determine the first voice of different users. It should be understood that voiceprint (voiceprint) is a sound wave spectrum that carries speech information displayed by an electroacoustic instrument. Because people’s vocal organs differ greatly in size and shape, the voiceprint atlas of any two people are different. Under normal circumstances, different people can be distinguished by voiceprint information or to determine whether the collected sound belongs to The voice of the same person. In this embodiment, the process of determining the first voice corresponding to the user by means of voiceprint recognition may be, for example, collecting the first voice of each user in advance, for example, breathing sound and/ Or snoring, and extract the characteristics of the first voice of different users, for example, obtain the short-term speech spectrum of different users’ voices; establish a voiceprint recognition model based on the characteristics of the first voice of different users; collect the first voice of the user during sleep Voice, and determine the first voice of different users according to the established voiceprint recognition model.

Exemplarily, the process of establishing a user's voiceprint recognition model may be, for example, obtaining the corresponding probability density function according to the characteristics of the user's first voice, such as a short-term speech spectrum, and establishing a Gaussian mixture model (GMM) The probability density of the spatial distribution is fitted with the weighted sum of multiple Gaussian probability density functions, so that the Gaussian mixture model can smoothly approximate the probability density function of any shape and obtain a parameter model that is easy to handle. Specifically, the parameter model obtained according to the probability density function of the short-term speech spectrum arranges the mean vector of each Gaussian component of the Gaussian mixture model together to form a super vector as a model of a certain user. According to specific parameter models corresponding to different users, the first voices corresponding to different users can be determined. Among them, the process of establishing different user parameter models through voiceprint recognition can refer to the existing process, which will not be repeated here.

S430: Acquire a power spectral density corresponding to the first sound according to the first sound.

Optionally, Fourier transform is performed according to the time domain signal obtained by the microphone to obtain the power spectral density of the first sounds of different users.

S440: When the power spectral density is lower than the first threshold, it is determined that the user has an apnea.

Optionally, when the power spectral density of the user's sleep sound is lower than the first threshold and reaches a second threshold, it is determined that the user has apnea, where the second threshold may be, for example, 10s or 15s.

Optionally, according to the number of times the power spectral density of the first voice of the different user is lower than the first threshold, the respiratory disorder AHI index of the user is determined. Specifically, the user’s AHI index is determined according to the number of times the user has apneas and the user’s sleep duration, and the AHI index can intuitively reflect the severity of the user’s sleep apnea.

As an example, when the power spectral density of the user’s breathing sound or the power spectral density of the snoring sound is lower than the first threshold, and the length of time lower than the first threshold reaches the second threshold, it is recorded as an apnea. , Record the total number of apneas (denoted as k), and calculate the sleep-disordered breathing coefficient according to its entire sleep duration (denoted as t), where AHI=k/t; or, record the user’s sleep per unit time The power spectral density of the sound is lower than the first threshold, and the number of times that the length of time lower than the first threshold reaches the second threshold.

For the manner of determining the sleep apnea status of the user according to the value of the AHI index, refer to step S330. To avoid repetition, details are not described herein again.

According to the method for apnea provided by the embodiment of the present application, the first voices of different users are collected through a microphone, and the voiceprint recognition algorithm is used to distinguish the corresponding sleep sounds of different users, which can achieve non-contact sleep apnea for multiple users Monitoring and risk assessment.

Fig. 5 shows a schematic flowchart of a method for apnea monitoring provided by an embodiment of the present application.

Wherein, the apnea monitoring method of the embodiment of the present application may further include the following steps.

S510: When it is determined that the user has apnea, the first device sends a wake-up message to the second device.

Among them, when it is determined that the user has apnea or the apnea has occurred for a certain period of time, the user is awakened.

Optionally, the second device may be a smart device surrounding the user, for example, a smart speaker, a smart light, a smart wearable device, a smart phone, etc., or it may be a smart device of the user's emergency contact, such as a smart phone.

Optionally, when it is monitored that the duration of the user's apnea has reached the third threshold, the first device sends a wake-up message to the second device, where the third threshold may be, for example, 20s.

S520: The second device wakes up the user who has experienced apnea by means of vibration, sound, light, etc.

Wherein, when the second device receives the wake-up message of the first device, it can wake up the user in different ways, for example, by means of vibration, sound, light, etc., to wake up the user who has experienced apnea.

Exemplarily, when it is detected through the first device that the duration of the user's apnea has reached the third threshold, the first device may send a wake-up call to linked devices around the user, such as smart speakers, smart lights, smart wearable devices, smart phones, etc. Messages enable peripheral smart devices to wake up users in time through vibration, sound, light and other corresponding methods. Wherein, the third threshold may be, for example, 20s or other values set with reference to a doctor's suggestion, which is not limited in this application.

Fig. 6 shows a device for monitoring sleep apnea provided by an embodiment of the present application. The device 600 includes a sound collection unit 610 and a data processing unit 620.

Wherein, the sound collection unit 610 is used to collect the first sound of at least one user.

Optionally, the data processing unit 620 is configured to obtain the power spectral density of the first sound according to the first sound.

Optionally, the data processing unit 620 may also be configured to determine the AHI index of the user according to the power spectrum density.

Fig. 7 shows another device for monitoring sleep apnea provided by an embodiment of the present application. The device 700 includes at least one microphone 710 and a processor 720.

Optionally, the microphone 710 is used to collect the first voice of at least one user.

Optionally, the processor 720 is configured to obtain the power spectral density of the first sound according to the first sound.

Optionally, the processor 720 may also be configured to determine the AHI index of the user according to the power spectral density.

A person of ordinary skill in the art may be aware that the units and algorithm steps of the examples described in combination with the embodiments disclosed herein can be implemented by electronic hardware or a combination of computer software and electronic hardware. Whether these functions are executed by hardware or software depends on the specific application and design constraint conditions of the technical solution. Professionals and technicians can use different methods for each specific application to implement the described functions, but such implementation should not be considered beyond the scope of this application.

Those skilled in the art can clearly understand that, for convenience and concise description, the specific working process of the above-described system, device, and unit may refer to the corresponding process in the foregoing method embodiment, and details are not repeated here.

In the several embodiments provided in this application, it should be understood that the disclosed system, device, and method may be implemented in other ways. For example, the device embodiments described above are only illustrative. For example, the division of the units is only a logical function division, and there may be other divisions in actual implementation, for example, multiple units or components can be combined or It can be integrated into another system, or some features can be ignored or not implemented. In addition, the displayed or discussed mutual coupling or direct coupling or communication connection may be through some interfaces, indirect coupling or communication connection of devices or units, and may be in electrical, mechanical or other forms.

The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the objectives of the solutions of the embodiments.

In addition, the functional units in each embodiment of the present application may be integrated into one processing unit, or each unit may exist alone physically, or two or more units may be integrated into one unit.

If the function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer readable storage medium. Based on this understanding, the technical solution of this application essentially or the part that contributes to the existing technology or the part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including Several instructions are used to make a computer device (which may be a personal computer, a server, or a network device, etc.) execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage media include: U disk, mobile hard disk, read-only memory (Read-Only Memory, ROM), random access memory (Random Access Memory, RAM), magnetic disk or optical disk and other media that can store program code .

The above are only specific implementations of this application, but the protection scope of this application is not limited to this. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed in this application. Should be covered within the scope of protection of this application. Therefore, the protection scope of this application should be subject to the protection scope of the claims.

Claims

A method for monitoring apnea, which is characterized in that it comprises:

Collecting the first sound of at least one user through the multi-microphone array of the first terminal, where each user is located in a different beam coverage area of the multi-microphone array;

Obtaining the power spectral density corresponding to the first sound according to the first sound;

When the power spectral density is lower than a first threshold, it is determined that the user has an apnea.
The method according to claim 1, wherein the collecting the first voice of at least one user through the multi-microphone array of the first terminal comprises:

Setting the pickup parameters of the multi-microphone array according to the positions of the first terminal and the user, so that different beams of the multi-microphone array cover different users;

The multi-microphone array collects the first voices of the users in different beam coverage areas respectively.
The method according to claim 1 or 2, wherein the obtaining the power spectral density corresponding to the first sound according to the first sound comprises:

Acquiring a time domain signal of the first sound according to the first sound;

The power spectral density of the first sound is obtained by using the time domain signal through Fourier transform.
The method according to any one of claims 1-3, wherein the method further comprises:

Pre-collecting environmental noise of the environment where the user is located, and determining the frequency bandwidth of the environmental noise;

Filter the environmental noise according to the frequency bandwidth of the environmental noise.
The method according to any one of claims 1 to 4, wherein the method further comprises:

Through the voiceprint recognition algorithm, the first voice corresponding to different users is determined.
The method according to any one of claims 1-5, wherein the method further comprises:

Determine the sleep duration of the user;

The AHI index of the breathing disorder of the user is determined according to the number of times the apnea occurs during the sleep of the user and the sleep duration.
A method for monitoring apnea, which is characterized in that it comprises:

Collecting the first voices of multiple users through the first terminal;

Determining first voices corresponding to different users according to a voiceprint recognition algorithm;

Obtaining the power spectral density corresponding to the first sound according to the first sound;

When the power spectral density is lower than a first threshold, it is determined that the user has an apnea.
The method according to claim 7, wherein the first terminal has a multi-microphone array;

The collecting first voices of multiple users through the first terminal includes:

Collect the first sounds of multiple users through the multi-microphone array of the first terminal, where each user is located in a different beam coverage area of the multi-microphone array.
The method according to claim 8, wherein the collecting the first sounds of multiple users through the multi-microphone array of the first terminal comprises:

Setting the pickup parameters of the multi-microphone array according to the positions of the first terminal and the user, so that different beams of the multi-microphone array cover different users;

The multi-microphone array separately collects the first voices of the users in different beam coverage areas.
The method according to any one of claims 7-9, wherein the obtaining the power spectral density corresponding to the first sound according to the first sound comprises:

Acquiring a time domain signal of the first sound according to the first sound;

The power spectral density of the first sound is obtained by using the time domain signal through Fourier transform.
The method according to any one of claims 7-10, wherein the method further comprises:

Determine the sleep duration of the user;

The AHI index of the user is determined according to the number of times the apnea occurs during the sleep process of the user and the sleep duration.
A device for monitoring apnea, which is characterized by comprising:

A sound collection unit for collecting the first sound of at least one user;

A data processing unit, configured to obtain the power spectral density of the first sound according to the first sound;

The data processing unit is further configured to determine that the user has an apnea when the power spectral density is lower than a first threshold.
The apparatus according to claim 12, wherein the obtaining the power spectral density of the first sound according to the first sound comprises:

Acquiring a time domain signal of the first sound according to the first sound;

The power spectral density of the first sound is obtained by using the time domain signal through Fourier transform.
The device according to claim 12 or 13, wherein the sound collection unit is further configured to pre-collect environmental noise of the environment where the user is located;

The data processing unit is configured to determine the frequency bandwidth of the environmental noise, and filter the environmental noise according to the frequency bandwidth of the environmental noise.
The device according to any one of claims 12-14, wherein the data processing unit is further configured to determine the first voices corresponding to different users through a voiceprint recognition algorithm.
The device according to any one of claims 12-15, wherein the data processing unit is further configured to determine the sleep duration of the user, and according to the occurrence of the apnea during sleep of the user The number of times and the sleep duration determine the AHI index of the user.
A device for monitoring apnea, which is characterized by comprising at least one microphone and a processor, wherein the device is used to execute the method according to any one of claims 1 to 11.